The changing epidemiology of VanB Enterococcus faecium in Poland

Increasing prevalence of VanB Enterococcus faecium in Polish hospitals reported to National Reference Centre for Susceptibility Testing (NRCST) prompted us to investigate the basis of this phenomenon. Two-hundred seventy-eight E. faecium isolates of VanB phenotype from the period 1999 to 2010 obtained by NRCST were investigated by multilocus sequence typing (MLST) and multilocus VNTR analysis (MLVA). Localization, transferability, and partial structure of the vanB-carrying Tn1549 transposon were studied by hybridization, PCR mapping, sequencing, and conjugation. VanB isolates almost exclusively represented hospital-associated E. faecium, with a significant shift from representatives of 17/18 lineage to 78 lineage after 2005. The vanB determinant, initially located mostly on transferable plasmids of the pRUM-, pLG1-, and pRE25-replicon types, later on was found almost exclusively on the host chromosome. Fifteen different plasmid and chromosomal insertion sites were identified, typically associated with single transposon coupling sequences, mostly not observed before. Our study demonstrates the significant change in the epidemiology of VanB-E. faecium in Poland, associated with the introduction and spread of the lineage 78 of the hospital-adapted E. faecium. These data point to the importance of the lineage 78 for the spread of vancomycin-resistance, determined by the vanB gene cluster, resulting in an increasing VRE prevalence in hospitals. This study also supports the scenario, in which representatives of the hospital-associated E. faecium independently acquire the vanB determinant de novo and spread within and among hospitals, concomitantly undergoing differentiation. Electronic supplementary material The online version of this article (10.1007/s10096-018-3209-7) contains supplementary material, which is available to authorized users.


Introduction
The importance of enterococci as etiologic agents of hospitalacquired infections (HAIs) is currently increasing [1], and common glycopeptide resistance among these bacteria is especially alarming [2]. Among two most ubiquitous van gene clusters, responsible for this phenotype, vanA confers resistance to both vancomycin and teicoplanin, and vanB typically determines resistance only to vancomycin [3]. The vanB cluster is predominantly associated with the Tn1549-type transposons [4], which may reside either on plasmids or on the bacterial chromosome [4][5][6][7][8][9]. During the initial steps of conjugative transfer of transposon, the staggered cleavage by the Int recombinase results in the formation of a circular intermediate, joined by a 5-to 6-bp sequence originating from the donor genome, termed a coupling sequence, which, after transposition, is found adjacent to the transposon termini in the recipient [7].
Among the two clinically most important enterococcal species, i.e., Enterococcus faecalis and Enterococcus faecium, the latter is particularly prone to the acquisition of antimicrobial resistance determinants, including vanA and vanB clusters (vancomycin-resistant E. faecium, VREfm), resulting in increasing proportion of VREfm among hospital E. faecium [10]. Concomitantly, an increase in the incidence of HAIs caused by E. faecium is observed [10,11], likely due to the selection and worldwide dissemination of successful hospitaladapted clonal complex 17 (CC17) [12] that combines resistance to several antimicrobials with the enrichment in pathogenicity factors and increased epidemic potential. The Bayesian Analysis of Population Structure (BAPS) of the data obtained by multilocus sequence typing (MLST) demonstrated that CC17 may be divided into two subgroups corresponding to major lineages 17/18 and 78 [13].
The vanB gene was identified in E. faecalis at the beginning of the 1990s [14]. In Poland, the first VREfm with vanB2 was isolated in 1999 [15] followed by a growing VanB prevalence in 1999-2005 [16]. A further increase in VanB-VREfm after 2005, noticed by the National Reference Centre for Antimicrobial Resistance and Surveillance (NRCARS), prompted us to investigate these important pathogens to better understand the factors underlying the spread of VanB-E. faecium in Poland.

Bacterial isolates and antimicrobial susceptibility testing
Altogether, 278 non-repetitive isolates with the VanB phenotype received by the NRCARS during 1999-2010 from 36 centers in 22 cities were investigated. Fifty-eight VanB isolates from 1999 to 2005 were partly characterized previously [16]; of these, 56 were available and 222 isolates were received in 2006-2010. Twenty-seven and 48 isolates were obtained from invasive and non-invasive infections, respectively, and 201 from carriage; for two remaining isolates, the source was not reported. Antimicrobial susceptibility was tested using the broth microdilution method [17] and the Etest method for vancomycin, teicoplanin, and daptomycin (bioMerieux, Marcy l'Etoile, France). Results were interpreted following the European Committee on Antimicrobial Susceptibility Testing (EUCAST)-approved breakpoints [18] and the Epidemiological Cut-Offs (ECOFFs) (http://mic.eucast.org/Eucast2/, 6th November 2017, date last accessed).

Analysis of Tn1549, insertion sites, and coupling sequences
The presence of int Tn1549 and ORF1 Tn1549 was confirmed by PCR, and the vanY-vanX sequence in Tn1549 was established using overlapping PCR and sequencing. The Tn1549 insertion sites were identified by inverse-PCR (iPCR) [26] with Bsp143I (Fermentas, Lithuania). Primers targeting sequences adjacent to Tn1549 were designed based on iPCR results. Sequences were analyzed with the Lasergene package (DNASTAR, MD, USA). Primer sequences are available upon request.

Statistics
The differences in distributions were evaluated by the chisquared test, with a p value ≤ 0.05 considered significant.

Antimicrobial susceptibility phenotypes and clonal relationships of VanB-VREfm in Poland
All isolates were analyzed by MLVA, yielding 23 different MTs; 13 non-typable isolates repeatedly yielded incomplete MLVA profiles ( Year

Diversity of the vanY-vanX region in Tn1549
All isolates were positive for ORF1 Tn1549 and int Tn1549 . Sequencing of the vanY-vanX region (encompassing genes vanY, vanW, vanH, vanB, vanX; Fig. 1a) revealed 26 variants among 57 isolates, representing all centers and STs within a center. The most numerous group included A1-A20 variants, differing only by single-nucleotide polymorphisms (SNPs) at 21 nucleotide positions and highly similar to the corresponding region in Clostridium spp. and Eggerthella lenta (Fig. 1b). The A-type variants were characteristic for 48 of investigated isolates from 31 centers, and associated with 14 STs and 16 MTs. The B variants differed from the A-type by several SNPs and 6-bp insertion between vanS-vanY. They were 99% identical to the variant reported for the V583 [33]. An insertion of the ISEfa11 between vanS-vanY in B-type yielded C variants (Fig. 1a). The D-, E-, and F-types represented probable derivatives of an A-type, with a deletion encompassing the nt 12-799 of vanW, an 11-bp deletion upstream vanY, and insertion of ISL3 between vanS and vanY, respectively. All vanB genes represented the vanB2 variant [34].

Analysis of Tn1549 insertion sites and coupling sequences
To determine Tn1549 insertion sites, selected isolates were analyzed by iPCR and thus obtained sequences were used to search GenBank and to design primers specific for a genetic neighborhood of Tn1549. These primers were used to screen the whole collection, revealing 15 insertion sites and 14 coupling sequences in total (Table 2). For two isolates, the coupling sequence could not be established due to the fact that sequences resulting from iPCR had no homologs in GenBank. Typically coupling sequences were identical in a given insertion site, with an exception of CS_P1a/CS_P1b in aacA-aphD and CS_P6a/CS_P6b in citH. The most prevalent coupling sequence, CS_C3 (198 isolates from 24 centers in 16 cities) was associated with 16 MTs and 14 STs. The first CS_C3 isolate was observed in 2006 (Table 1B).
Analysis of Tn1549 localization, plasmidome composition, and vanB transferability Seventy-eight isolates were analyzed by S1/PFGEhybridization with the vanB probe (Table 1 and Supplementary Fig. 1). These isolates represented all observed variants of coupling sequence and hospital centers; additional isolates from the same center were included in the case of isolates showing plasmid localization of vanB. In the case of 39 isolates with coupling sequence C1-C8, vanB hybridized with a band of high-molecular weight, consistent with transposon insertion within chromosomal sequences and 39 isolates showed hybridization with plasmids from3 0 to~310 kb in size; in five isolates, vanB was located on two plasmids. These hybridization studies and iPCR/PCRbased analyses of coupling sequences were consistent with the chromosomal localization of Tn1549 for 227 isolates (81.6%) and plasmid localization for 50 isolates (18.0%); in a single case, a presumable integration of plasmid into chromosome was observed (variant CS_6a1 from WAW2). Isolates with the plasmid localization of vanB were much more prevalent among early VREfm, i.e., from 1999 to 2005 (61% of these isolates) compared to the isolates collected from 2006 to 2010 (0.7% of these isolates, p < 0.001). Among isolates with the plasmid localization of vanB, rep17 pRUM was found among 49 isolates, followed by rep pLG1 , rep2 pRE25 , rep1 pIP501 , rep pMG1 , and rep9 pAD1 (42,35,32,27, and 5 isolates, respectively). Thirty-one and 12 of these isolates carried axe-txe and ω-ε-ζ), respectively. S1-PFGE/hybridization analyses revealed that 29, 23, and 22 plasmids hybridized with the rep pLG1 , rep17 pRUM , and rep2 pRE25 probes, respectively (Table 1A). In several cases, a single plasmid was associated with two or three rep genes. Sixteen vanB-plasmids hybridized with the axe-txe probe; among them, 13 co-hybridized with rep pLG1 and 12 with rep17 pRUM , respectively. Nine plasmids did not hybridize with any of the four probes used. Sequencing revealed a low variability of rep genes within this group that included two, one, three, one, and three variants of rep1 pIP501 , rep2 pRE25 , rep17 pRUM , rep pMG1 , and rep pLG1 , respectively. Among 50 isolates with the plasmid localization of vanB, 43 isolates (86.0%) were able to transfer vancomycin resistance while conjugation experiments involving 32 representative isolates with various chromosomal localizations of vanB were negative in 29 cases.

Discussion
The first VanB-VREfm was detected in Poland in 1999 [15], and our study investigated the VanB epidemiology during the following 12 years. Considering a relatively moderate incidence of VREfm in Poland during this period (e.g., in 2010 amounting to 7.8% of invasive infections [http://ecdc.europa.eu/en/publications/Publications/1111_ SUR_AMR_data.pdf.pdf; 6th November 2017, date last accessed]), it may be assumed that our collection reasonably well reflected the epidemiological situation in Polish hospitals. Although initially VanA represented the major VREfm phenotype in Poland [16,35], after 2006, the NRCARS recorded an increasing number of VanB-VREfm, affectin g sev era l h osp itals. T he c urre nt glob al epidemiology of VREfm shows considerable differences, with VanA predominant in Europe and the USA [36], and VanB constituting over 80% of invasive VREfm in Australia [37]. A recent rise of VanB-E. faecium has been reported in Germany [8]. Nearly all isolates in our study belonged to the hospital E. faecium, since 2006 with the predominant role (89%) of lineage 78. VanB-VREfm belonging to this lineage were responsible for recent outbreaks in Germany, Sweden, and Australia [38], and representatives of lineage 78 played a role in vanA dissemination in Polish hospitals [34].   While the structure of Tn1546, harboring vanA shows a high variability [34,39], the vanY-vanX region in the vanB gene cluster appeared to be less divergent. In the studied collection, the A-type showed the highest prevalence, with variants very similar or identical to these found in E. faecium in Australia, France, and Taiwan [40][41][42], and in the pMG2200 plasmid of E. faecalis [5]. Importantly, the A variants are also present in gut anaerobes such as Clostridium spp. and E. lenta [26,40], a presumable reservoir of Tn1549-type transposons. Genomic analyses of VanB-E. faecium and concomitantly isolated vanB-positive gut anaerobes indicated the epidemiological significance of de novo acquisition of Tn1549 by hospital-adapted E. faecium [7,41]. The B-type characteristic for the first vancomycin-resistant E. faecalis V583 strain [33], to our knowledge, has not been reported in E. faecium so far. The presence of ISs targeting the vanS-vanY intergenic region (resulting in C-and F-types), was observed also elsewhere [43]. Such variability of vanB clusters may be useful in analyses of suspected VRE outbreaks. For example, plasmid-located Dtype was found in isolates representing various MTs and STs from the WAW1 hospital (Table 1A). Thus, a spread of a stablẽ 150 kb conjugative plasmid of undetermined replicon type, harboring this specific variant of the vanB cluster was most likely responsible for the outbreak. Similarly, although isolates from SZC2 differed both in the clonal composition and vanB-associated plasmidome, C-type was detected in all these isolates (Table 1A), indicating extensive plasmid recombination during an outbreak. Until now, more detailed knowledge of plasmids carrying vanB in E. faecium remains limited [38]. In our study, vanB-plasmids represented mostly the rep pLG1 , rep17 pRUM , and rep2 pRE25 replicons, similarly to the situation observed for vanA-plasmids in Poland [34]. The original pLG1 contained the complete vanA gene cluster [44] and plasmids with this rep were responsible for an increase of HLGR among E. faecium in Norway [45] but, to our knowledge, vanB-plasmid of the rep pLG1 type has not yet been reported. The second observed rep type, rep17 pRUM was involved in a multicenter VanB outbreak in Sweden [46] and in the HLGR spread in Norway [45]. The rep pLG1 and rep17 pRUM genes frequently occurred together and in combination with the axe-txe, characteristic for plasmids with these replicons [34,45,46]. Plasmids harboring vanB were typically transferable by conjugation and during outbreaks (e.g., in KRA2, WAW1 and SZC2) were associated with diverse clonal backgrounds. Such plasmid dissemination was additionally accompanied by presumable recombination/co-integration events, resulting in the observed variability of vanB-plasmids. A similar dynamics was observed also for rep17 pRUM -type vanA-plasmids [30]. Recombination/ co-integration likely contributed to the association of vanB with more than a single rep, observed in the current study and characteristic for E. faecium plasmids in general [30,34,38,45]. Two plasmid-located genes, aacA-aphD and citH, showed the integration of Tn1549 with different coupling sequences and might represent transposon integration hotspots. Such hotspots were indeed observed for E. faecium [7].
Isolates with plasmid-borne vanB prevailed until 2006, and later this determinant showed usually a chromosomal localization. This change occurred in parallel with the emergence and spread of lineage 78. The predominantly chromosomal localization of the vanB cluster in lineage 78 was observed recently also in Germany and Australia [7,8,40]. Two variants of coupling sequences, CS_C1 and CS_C3, were associated with two most numerous groups of isolates (Table 1). Twenty-one isolates with the CS_C1 variant, present in 17/ 18 lineage and A1 type of the vanY-vanX region, showed multicenter distribution over 2003-2008. These isolates showed some divergence of their STs/MTs, which may be explained by a transfer of transposon-containing region to a new clonal background [8] and/or exchange of other genomic regions, leading to formation of new STs/MTs [9]. An even more complex epidemiological situation was associated with isolates harboring the CS_C3 variant. This particular group appears to be the main contributor to the increasing proportion of VanB among VREfm and general increase of prevalence of VRE in Poland and was responsible for extensive outbreaks, e.g., in KSZ and POZ2 hospitals. With the exception of a single isolate with CS_C5, which shared a coupling sequence and insertion site with several ST192 isolates from Germany [8], none of the remaining coupling sequences showed identity to coupling sequences described elsewhere [7][8][9]. This finding is consistent with proposed independent de novo acquisition of Tn1549 [7].
This study provides an analysis of VanB-E. faecium, performed on a country level and over an extensive period of time. We demonstrate a significant change both in the clonal background as well as localization of Tn1549-type transposons, carrying vanB genes. Our study supports the role of lineage 78 of the hospital-adapted E. faecium, presumably acquiring de novo the vanB determinant, followed by spread and differentiation of certain strains as a major factor beyond the current increasing prevalence of VanB-VREfm in Polish hospitals. statutory funds from the Ministry of Science and Higher Education, Poland.

Compliance with ethical standards
Isolates were obtained as a part of routine activity of the NRCARS and were analyzed anonymously in a retrospective manner. Ethical approval and informed consent were thus not required.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.